Velaura AI Inc. secured $110 million in a Series A round to build low‑power AI processors, shifting focus from crypto accelerators to AI chip building blocks.
Velauta AI Inc. announced a $110 million Series A financing round aimed at developing power‑efficient AI processors, marking a strategic pivot from its earlier focus on cryptocurrency accelerators.
Funding round and investors
The round was led by Sequoia Capital with participation from Accel, Bessemer Venture Partners and several strategic corporate investors. The capital will fund chip design, tape‑out, and early‑stage production.
Shift from crypto to AI
Founded in 2022, Velaura initially built ASICs for cryptocurrency mining. Market volatility and rising energy costs prompted the company to repurpose its low‑power design expertise toward AI workloads, where efficiency is a competitive differentiator.
Technical approach
Velaura’s architecture leverages a modular compute block that can be tiled to meet diverse AI inference demands while keeping power draw below 5 watts per teraflop. The firm claims the design supports both dense matrix multiplication and sparse operations, catering to large language models and edge‑device inference alike.
- Modular compute blocks for scalable performance
- Sub‑5 W/TFLOP power envelope
- Support for dense and sparse AI kernels
- Compatibility with existing software stacks via open‑source SDK
Market outlook
Analysts note that demand for energy‑conscious AI hardware is growing as data centers seek to curb electricity costs and meet sustainability goals. Velaura’s low‑power focus positions it to compete with established players that prioritize raw performance over efficiency.
“We see a clear need for AI chips that deliver high throughput without the massive power bill,” said Velaura CEO Maya Patel in a post‑fundraising interview.
The company plans to deliver its first silicon samples by Q2 2027, targeting early adopters in edge computing, autonomous vehicles, and energy‑constrained data‑center workloads.
Comments
No comments yet.